Troubleshooting a vehicle often requires some form of customer service; to help guide a customer towards a resolution. Consumer reporting of faulty vehicular components can be made via a telephone- based service call, or gathered through telemetry via embedded intelligent transportation systems (ITS). During data transmission, free text is created in the dialogue between the customer and the service representative. While free text is generated, key details of the discussion can be extracted and recorded. This paper describes methods used to process the recorded free text data. This method can help classify and direct the call to the correct channel of support tools and resources. An anonymous customer service report consisting of 75,000 calls was used for feature extraction. Five thousands of the calls were used in supervised learning to support data classifications. The matrix of data was evaluated for accuracy and repeated to minimize error and increase accuracy. By incorporating hand-crafted features using domain expertise, our natural language processing (NLP) based approach can achieve 85 percent accuracy in classifying the service calls.
Mining Vehicle Failure Consumer Reports for Enhanced Service Efficiency
01.09.2019
297413 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Vehicle Consumer Complaint Reports Involving Severe Incidents: Mining Large Contingency Tables
Transportation Research Record | 2018
|Rules mining on hybrid electric vehicle consumer complaint database
Taylor & Francis Verlag | 2023
|Service Please: Consumer viewpoint
British Library Online Contents | 1999
Mining and Classifying Aviation Accident Reports
AIAA | 2019
|Service brake of mining rubber-tyred vehicle dead axle
Europäisches Patentamt | 2016
|